Abstract: A color image of an inspection object is taken by an imaging means capable of taking a color image to obtain color information of an RGB color space. A gray-scale image of a color component of the RGB color space or another color space is generated, and the inspection object is detected by a pattern recognition technique. Alternatively, a binary image is generated from the generated gray-scale image, and the inspection object is detected by performing pattern recognition on the binary image. Color data of a pixel occupied by the detected inspection object is compared with color data of a non-defective inspection object which is previously prepared to judge whether or not the inspection object is defective. In addition, this judgment result is reflected in another manufacturing step through a network and product quality is improved.

Abstract: A color image of an inspection object is taken by an imaging means capable of taking a color image to obtain color information of an RGB color space. A gray-scale image of a color component of the RGB color space or another color space is generated, and the inspection object is detected by a pattern recognition technique. Alternatively, a binary image is generated from the generated gray-scale image, and the inspection object is detected by performing pattern recognition on the binary image. Color data of a pixel occupied by the detected inspection object is compared with color data of a non-defective inspection object which is previously prepared to judge whether or not the inspection object is defective. In addition, this judgment result is reflected in another manufacturing step through a network and product quality is improved.

Abstract: A color image of an inspection object is taken by an imaging means capable of taking a color image to obtain color information of an RGB color space. A gray-scale image of a color component of the RGB color space or another color space is generated, and the inspection object is detected by a pattern recognition technique. Alternatively, a binary image is generated from the generated gray-scale image, and the inspection object is detected by performing pattern recognition on the binary image. Color data of a pixel occupied by the detected inspection object is compared with color data of a non-defective inspection object which is previously prepared to judge whether or not the inspection object is defective. In addition, this judgment result is reflected in another manufacturing step through a network and product quality is improved.